Sunday, March 20, 2005
Ontology streaming as an enabling technology
A gap of almost 11 months
occur before the next bead is posted in this tread à [29]
Communication to the BCNGroup from Nan Gelhard
I've put together some thoughts on Streaming Ontology. Please share them as appropriate. Ask our colleagues to share his/her comments and I will try to clarify my ideas.
A few years ago I wrote some rules for e-Commerce. Consider these thoughts to be an extension of those requirements.
Best, Nan
Querying the commerce data set (merchandise, customer service, shipping information, price and availability, etc.) should be as comfortable as talking to a neighborhood human being.
I specify neighborhood human because language is locally variable. Note that though it is easier to talk to a person than to a CPU, human conversation is not trouble-free. Some subjects/transactions require iterations.
The more esoteric the data set the more negotiation -- explanation -- may be required to make a transaction or get information. A virtual shopping mall, like e-bay or Amazon, serves a customer by authenticating the seller for the buyer, and in Amazon's case, by securing the money transfer. I understand that Streaming Ontology can give me access to unstructured data; but can it support transactions?
How trustworthy are the query results when the info is written numerically but the meaning is nominal? For instance, could an ORB parse 351W as an engine, not a width? Credit card numbers are structured, but manufacturers' part numbers are mostly random. How much of a legacy database structure can it leverage?
One of the biggest hurdles to getting to e-Market is converting or building info sets that are accessible to customers. Making sense of, and giving customers access to, unstructured data could shorten the time to market.
If I have a niche product line, and you can give me access to a world market, even though my share is small, my sales can be huge. That is stating the obvious; but I don't think everyone noticed what Amazon did to the out-of-print book market.
I credit Amazon with saving independent booksellers. First, it significantly grew the market for out of print books by telling people there was out of print inventory and making it accessible. Amazon also changed what was the extant market. There is no longer a market for first editions. Once Amazon provided a mechanism for sellers to display to buyers across the country, collectors found that first editions were only geographically rare. It also turns out that book collectors are only geographically rare. Amazon created a community of interest for the sellers to market to.
Amazon's new bookstore functioned as the anchor store in the mall. Notice that though you used to have to specifically ask to see books for sale by Amazon affiliates, now they are presented as part of all search results.
Amazon leveraged the precisely structured and standardized ISBN numbers and the Library of Congress book info to its great advantage. Can ORBs help me, a small retailer with no particular computer database knowledge a way to make my wares available online? What about a chain store, with or without an e-Commerce system?
Can Streaming Ontology leverage my existing somewhat structured data sets?
Can you use ORBs to reveal or create communities of interest?
Another hurdle for profitable commerce, especially for products with waxing and waning popularity, is good business intelligence about rising stars and failing fashions. By analyzing sales of the current product line, I can direct my advertising effort and control inventory -- to a point. Can you give me access to information about consumer attitudes about products that aren't in my line? In other words, can a Streaming Ontology show me opportunities?
As I understand the virtual Mall’s business model, its value to consumers is:
1) Presenting an interesting collection of merchandise, presumably from many retailers, with a single, secure payment system.
And its value to retailers is:
1) Create traffic to the virtual mall.
Amazon and now e-bay are leveraging the brick and mortar mall model by partnering with established retailers. Consider that many people understand what Sears has to sell better than what Tibetan Gifts does. Amazon adds a Gap store to its e-Mall and some of the customers that attracts visit the boutiques.
To be a success, any virtual mall needs to let consumers know what kind of stuff its stores have for sale. Are Streaming Ontology answers available to search engines? Are the answers relevant enough to function as advertising?
(Note that you know what Gap sells by virtue of its branded advertising.)